Josy Xi Zhou
The University of North Carolina at Chapel Hill, USA
Title: The Combination of statistical methods in selecting potential covariates from a large number of observed factors based on the structural relations for epidemiological studies
Biography
Biography: Josy Xi Zhou
Abstract
Many researchers summarized different methods of selecting covariates (e.g., forward and backward selection, high-demission proxy adjustment) to reduce bias in epidemiological studies. However, as the lack of a combinational statistical method and an effective programming perform for covariates selection, misclassifying, misadjusting, or missing covariates on fixed treatment comparisons or predicted the outcome is still a primary issue. For example, some covariates that are predictive of the outcome but have no influence on treatment status are often adjusted to balance treatment. As known, adjustment for such covariates can increase bias. The study conducts an essential algorithm called “Combination of Statistical Methods (CSM)” regarding the relation of variables to address the complex issue of limiting the number of potential covariates to a large number of covariates for adjustment in order to compare the causative effects of treatments.
CSM by standardized difference, bias, and odds ratio to declare the association of each covariate and select confounding covariates from a large number of observed variables (i.e., demographics, empirical, baseline-identifier, and measured potentially hidden covariates during the study window) and model adjustment as well. It is neither limited sample size nor the number of covariates. Moreover, CSM can further help researchers to discover important confounders that the previous studies ignored. In a word, CSM provides an efficient, convenient, and timesaving computing platform on selecting covariates for adjustment based on the association of each covariate in a summarized table, and generates effects or the evidence that limitation of adjusted preferred covariates in empiricism.